Singxer SU-1 and Linux - Producing "white noise" when changing tracks

Eric,

I think I’ve cracked it. If you look up my log from about 15-5mins ago you might see what’s happening. Long story short, I was trying to isolate the issue and I had thought this was something to do with going from 16bits to 24bits tracks. But my experiments proved this was not the case. So I then set the DSP to resample to 192khz and that didn’t fix the issue so it’s not related to switching sampling rates. I then thought it might be to do with file formats but flac to MP3 or vis versa didn’t seem to evoke the issue and to be honest I can’t see how that would affect anything anyway.
So I started playing around with other things inside the device settings, Under DSD Playback Strategy, I changed “native DSD” to “convert to PCM”. Result = issue solved (at least as far as I can see).
What is odd is that the files that are having the issue are not DSD files at all. In fact, it’s almost the polar opposite in that MP3 files seem to give the most issues - pretty far removed from a DSD file!

Example, switching from Panthu Du Prince FLAC 24bits 44.1khz to Peter Selway “coming up for air” MP3
44.1khz 24bits 260kbps causes the white noise and dropout every time when DSD is set to “native”. Switching from the Peter Selway MP3 track back to the Panthu Du Prince track does not cause the issue. Both files are 44.1khz, 24 bit and the only difference is the file format and I guess data-rate is lot higher from the FLAC file.

So to conclude - I can now set my Singxer to 32bit operation and can avoid dropouts when playing any type of music from any format / bitdepth / dampling rate at the moment - as long as I keep DSD playback strategy to “convert to PCM”. Weird!!!

Mark

Hi,

Picking up on @Mark_Kelly1 point I’ve been switching between Euphony OS and ROCK on my Intel NUC. When I run Roon Core on Euphony the driver for the Singxer does not support native DSD, whereas in ROCK it does. However I still get the white noise issue when switching from 16 to 24 bit but much less frequently as I need to have DSD Convert to PCM enabled.

It always occurs when stepping down from 24 to 16 bit on ROCK. It occurs frequently when going up to 24 from 16 on ROCK.

Again as suggested by @Mark_Kelly1 keeping everything at 16 bit stops the issue completely.

Given that this issue doesn’t occur in Windows or Mac (allegedly) I’m still suspicious of the UAC2 driver in various flavours of Linux.

@eric I would be happy to let you switch logging on but as I say I’m bouncing between ROCK on my M2 SSD and Euphony OS on a “Live CD” USB stick. There is a difference in sound characteristics but I don’t know which one I prefer…

I think Euphony is based on Arch Linux as is ROCK?

Kind Regards,

Ian.

@Mark_Kelly1 and @Ian_Oliver ---- Thank you both for the follow ups and the diligence here :clap: Having this additional feedback is greatly appreciated!

Mark I enabled diagnostics on your account and can confirm that the new report has hit our servers. Ian I am going to do the same for you so I can attach the report to your ticket for our techs to review.

Furthermore, I did talk to our tech team regarding your latest posts and they have asked if you could please confirm what firmware version is being used on the SU-1?

-Eric

@eric I’m using file 2.22 which is version 2.02 of the firmware for the PS Audio DAC. My understanding of 2.00 vs 2.02 is the former swaps Left & Right channels for the Holo Audio Spring DAC.

I was running 2.02 up until Tuesday morning.
I upgraded the firmware to 2.22 to make sure that this was not the source of the issue and it is still there.
So both the new and old firmware have the same issue.

@Ian_Oliver and @Mark_Kelly1 — Thank you both for getting back to me with the requested information!

Both of your tickets have been updated with the above and diagnostic reports from both of your systems have been attached as well. I am going to be passing over to our tech team for review and once I have an update I will be sure to share the team’s thoughts/findings with you both asap.

-Eric

Friday 19th October 09:47 AM local UK time

Going from 16bit 44.1 to 24bit 44.1: George Michael, Songs from the Last Century, The First Time Ever I Saw Your Face to Rag’n’Bone Man, Human, Human

Friday 19th October 10:25 AM local UK time

Going from 16bit 44.1 to 24bit 44.1: Annie Lennox, Diva, Primitive to Amy Winehouse, Lioness: Hidden Treasures, Valerie ['68 Version]

@eric

So bit depth change from 16 to 24 on both occasions but sample rate of 44.1 remained consistent.

Hi @Ian_Oliver ---- Thank you for providing the requested time stamps, VERY appreciated!

I know you mentioned that you had been switching back and forth between various devices acting as core machines. Just to avoid any assumptions :innocent: this was performed on your ROCK machine correct?

-Eric

@eric Yes ROCK on NUC.

Thanks for verifying @Ian_Oliver! I am going to enable diagnostics on your account now and as mentioned in my previous post this action will automatically be generated/uploaded a diagnostics report containing a set of your Roon logs do the next time Roon is active on that machine. Once the report arrives I will be sure to touch base so you know that I have it.

-Eric

@eric replicated same problem using same songs starting at 15:39 UK time

When changing bit depth AND sample rate i.e. 16 to 24 and 44.1 to 96 kHz there doesn’t seem to be a problem. However when sample rate remains 44.1 but bit depth goes up from 16 to 24 then we ALWAYS trigger white noise.

1 Like

Great! The diagnostics report has been received @Ian_Oliver. Thanks again!

-Eric

Any updates on this?

If I can add a data point, the same behaviour described in this thread occurs when using a NuPrime uDSD. Soon as there’s a switch in bit depth or sampling rate there’s a short burst of white noise and then it’s unresponsive until it’s power cycled. I would not be surprised to find they use the same USB chip.

Same here on Arcam irDac ii and looks like other DACs have same issue. There is an alarming amount having the same issues.

I think the nuprime has an XMOS USB input. And your Arcam IrDAC 2 has one as well going by this picture: http://www.aq.cz/images/d5_CatalogueModule-Files/243-file-IR_DACII_Hifi_World_Aug_16.pdf

A pattern is forming…

Yep I think support are now aware of the issue with the xmos as it was mentioned on another thread with user having same issue.

Worth pointing out though that in the case of the uDSD the issue is specific to Roon. I’m told by other uDSD users they do not experience same using eg jriver.

Yes same with my Arcam mpd no issues at all, it’s just Roon that has the issue.

Hi @eric

For the past year or so, my digital chain has been plagued by this same white noise issue that @Mark_Kelly1 and @Ian_Oliver have narrowed down to this XMOS chip (great work guys, and THANK YOU). My chain:

Windows 7 64-bit laptop, i7-2640M processor, 8GB RAM running Roon Core
Content stored locally on HDD and streaming Tidal

NUC7PJYH endpoint running Roon Bridge on AudioLinux, USB output
Uptone Audio ISO Regen
Singxer SU-1 (FW v2.2) outputting I2S
Holo Spring KTE DAC

Note: this issue is also present when using a Sonore ultraRendu as an endpoint (same chain otherwise)

Using Ian’s method, I am able to consistently trigger a crash in the Singxer when switching back and forth between 16 and 24 bit files, both sampled at 44.1kHz. Mark’s DSD playback strategy tweak isn’t effective with my system, as I tested each of the following scenario’s 5 times apiece:

DSD playback strategy - Native
playing 16-bit file, then switching to 24-bit: crashed 3 out of 5 times
playing 24-bit file, then switching to 16-bit: crashed 3 out of 5 times

DSD playback strategy - Convert to PCM
playing 16-bit file, then switching to 24-bit: crashed 3 out of 5 times
playing 24-bit file, then switching to 16-bit: crashed 2 out of 5 times

To reset, I have been power cycling the Regen (and therefore the Singxer’s bus powered USB input) and everything returns to normal.

I have not seen any crashes when setting Roon DSP to upsample everything to a consistent rate. However, my preference at the moment is listening to native file playback, so I wanted to pile on in hopes of getting this resolved.

This thread seems to have gone inactive lately so please let me know if there’s any news.

Dan